I have developed a Python Notebook that evaluates probabilities for observing Transiting Planets within the Rubin/LSST framework. It does no file accessing other than writing some output files. But, it does a lot of computing. I initially developed it on my own Mac desktop and have transferred it to the lab. I first ran it with 2 CPU’s and then with 4 CPU’s. Even with 4 CPU’s, it’s run time is no better than my Mac. Is there a feature or configuration that makes available more processing horsepower within the Lab? By the way, the normal run time on my Mac is about 30 hours. If it would help, I could split the work into multiple notebooks and submit them for processing simultaneously. Would that be better?
The RSP is not a supercomputer or batch computing system - its purpose is to offer high-bandwidth low-latency access to Rubin’s (imminent!) massive data holdings - the compute available to the sizeable data rights holding community, at no cost, is indeed equivalent to a (quite modest in fact) laptop.
If you are not doing data access (which it sounds like you are not) you are not taking advantage of the Rubin Science Platform’s prime capability and for many people their own laptop is indeed going to be a more performant choice in such a case.
1 Like
Thanks.